How to configure neural search for two fields

OpenSearch
1

I have a simple database of articles (title + body):

"mappings": {
    "properties": {
        "title": {
            "type": "text",
            "analyzer": "..."
        },
        "body": {
            "type": "text",
            "analyzer": "..."
        },
    }
}

When searching I use a multi_match query to boost the title field:

"query": {
    "multi_match": {
        "query": "...",
        "fields": ["title^1.2", "body"],
    }
}

I would like to add semantic search here (hybrid).
What is the best way to implement this behavior using neural search? I’ve read the tutorial, but still can’t get what to do with the second field. If I calculate two vectors (title_embedding, body_embedding) how can I combine them in a neural query?

One more question: without boosting. Is there any way to concatenate two fields for ingest pipeline to avoid making another one in the mappings section? Something like this:

"mappings": {
    "properties": {
        "title": {},
        "body": {},
       "title_body_embedding": {} // vector of title + ' ' + body
    }
}

PS: sorry, my English is not good